Machine Learning-Assisted Beam Alignment for mmWave Systems

• DOI: 10.1109/tccn.2021.3078147
• Publication Date: 2021-05-07T21:08:09Z
• Authors: Yuqiang Heng; Jeffrey G. Andrews
• Keywords: Extremely high frequency

Beam alignment is a challenging and time-consuming process for millimeter wave (mmWave) systems, particularly as they trend towards higher carrier frequencies which require ever narrower beams. We propose beam method that assisted by machine learning (ML), where we train ML models to predict the optimal access point (AP) – or best few candidates user equipment (UE) given just its GPS coordinates, can be fed back UE estimated network using emerging localization techniques. evaluate with data generated state-of-the-art commercial ray-tracing software in realistic urban topology. Even dynamic scatterers imperfect our proposed greatly reduce search space (number of candidates) finding AP beam. For example, 28 GHz scenario 64 beams, reduces approximately 4x selection over 10x while achieving 95% accuracy. provide dataset ease reproducing extending results, suggest coupled suitably trained significantly speed up current procedures.